™ Vispera Shelfsight Automates Shelf Inspection for Optimal Inventory Management Smart inventory tracking optimized by the Intel® Distribution of OpenVINO™ toolkit “The results from our exercise in Intel® DevCloud for the Edge show that neural network computing performance on a CPU now matches a GPU. These performance benchmarks are constantly improving, thanks to new, ...innovative tools. Now, advanced edge inference solutions like ours are feasible with CPUs, allowing us to run our solution as a real-time IoT application operating on Intel® Xeon® processors at multiple stores.” — Erdem Yoruk, Chief Scientist, Vispera Information Technologies Up to % 20 2 improvement in planogram compliance Real-time shelf monitoring makes inventory management faster and more accurate For retailers, customer experience often hinges on maintaining optimum inventory levels. Historically, managing stock has been a time-consuming task relying on floor personnel manually inspecting each shelf, with stockouts costing retailers an 1 estimated $1 trillion per year. Now, Vispera Information Technologies brings the power of deep learning and computer vision to store shelves, increasing product 2 availability while improving planogram compliance by up to 20 percent. Vispera’s new Shelfsight™ solution automates the shelf inspection process, maintaining a constant digital record of shelf content throughout the store. Using Intel® Xeon® Scalable processors optimized with the Intel® Distribution of OpenVINO™ toolkit and Intel® DevCloud for the Edge, Shelfsight™ drives smart optimization of shelf execution, detecting out-of-stock (OOS), misplaced, and excess items as well as other noncompliance issues. With a better grasp of shelf stock, retailers can make data-driven decisions about inventory purchases and allocation of store personnel. Challenges: Fast-moving inventory and multivendor optimization Shoppers have high expectations for finding the products they want, when they want them. Although out-of-stock items contribute to poor customer experience and lost sales, understanding shelf inventory well enough to correctly identify specific OOS items requires more than generic object recognition. Visual inspection of shelf stock requires recognition of many fine-grained products. Modifications to product appearance, introduction of new products, and delisted products create a constantly evolving inventory landscape that must regularly be updated. Retailers need a turnkey, fixed-camera, shelf-monitoring solution to meet their challenges, using an orchestrated multivendor approach, including hardware suppliers for cameras and on-premises servers and system integrators. Real-time execution of deep neural networks requires optimized models that maximize CPU resource efficiency. Solution Brief | Vispera Shelfsight™ Automates Shelf Inspection for Optimal Inventory Management Solution: Real-time on-demand shelf inspection at Shelfsight™ automates the shelf inspection process, using shelf-facing fixed cameras in real time, continuously, and 2 with near-perfect accuracy. Using state-of-the-art deep learning algorithms developed by Vispera and optimized with Intel Distribution of OpenVINO toolkit and Intel DevCloud for the Edge, Shelfsight™ detects and recognizes individual products with stock keeping unit (SKU)–level granularity, rather than generic product types. The solution also detects price tags, reads them, and matches them with the nearest product. Shelfsight™ maintains a complete digital model of shelf content, generating guided notifications with replenishment tasks for store personnel. Noncompliance can be rapidly addressed with fast action to eliminate OOS items and planogram errors, maximizing sales by ensuring that customers can find the products they’re looking for. Shelfsight™ can provide measurements in two modes: scheduled or on demand. Scheduled mode takes measurements at a preset frequency—10 minutes for stores currently using the system. With each measurement, Shelfsight™ detects out-of-stock events and noncompliance issues, then issues push alerts to store personnel using in-store screens. For a typical supermarket with 100 m of Read the full Solution Brief AI and Computer Vision Retail Inventory Tracking.